Pragmatics 20:3.425-447
(2010)
International Pragmatics Association
ETHNICITY IN LINGUISTIC VARIATION:
WHITE AND COLOURED IDENTITIES IN AFRIKAANSENGLISH CODE-SWITCHING
Gerald Stell
Abstract
The Afrikaans speech community is characterized by a long-standing rift between Whites and Coloureds,
and is for a large part bilingual, with English being increasingly integrated in its stylistic repertoire. Yet,
the history of English is different across the White/Coloured divide, as in particular in terms of diffusion
and in terms of ideological associations. The question we wish to ask is twofold. First, how far may there
be a question of ethnic norms of Afrikaans-English code-switching? Second, if norms of code-switching
are different across the ethnic divide, is code-switching used differently in the negotiation of White and
Coloured identities?
This contribution is organized in three main parts. First, we give an overview of the different
norms of Afrikaans-English code-switching encountered across Whites and Coloureds on the basis of a
corpus of informal speech data. Then we give an overview of the sequential patterns of Afrikaans-English
code-switching following a CA methodology. Finally, we determine with the help of macrosocial
knowledge in how far these different forms and functions of Afrikaans-English code-switching are made
relevant to the projection of White and Coloured identities in South Africa’s current post-Apartheid
context on the basis of select individual examples.
The results of our analysis indicate that Afrikaans-English code-switching in the Coloured data
displays the features of a ‘mixed code’, which is perceived as a ‘we-code’, where English input tends to
be stylistically neutral. By contrast, English input is more syntactically and sequentially salient in the
White data, and more visibly serves purposes of identity-negotiation. Despite those differences, there
remains a clear correlation in both White and Coloured samples between the use of English monolingual
code and affiliation with ‘New South African values’.
Keywords: Code-switching; Afrikaans; Sociolinguistics; Conversation analysis.
Introduction
The status of ethnicity with regard to linguistic variation has received considerable
attention, but the theoretical perspectives from which it has been approached are not
necessarily compatible. Whereas in variationist approaches ethnicity has often been
implied in sets of variants of which the co-occurrence is presented as predictable on the
grounds of social and contextual variables, approaches influenced by social
constructivism have presented ethnic identity as a negotiated construal, projected in
interactions by means of socially meaningful stylistic choices. Of interest here is the
expression of ethnicity through linguistic variation involving varieties which are largely
shared, yet attached historically different symbolic connotations across the ethnic
divides of a given speech community.
426
Gerald Stell
The Afrikaans speech community provides a case in point: It is characterized by
a long-standing rift between Whites and Coloureds, and is for a large part bilingual,
with English being increasingly integrated in its stylistic repertoire. Yet, the history of
English is different across the White/Coloured divide, as in particular in terms of
diffusion and in terms of ideological associations. Nowadays, English is the
unchallenged lingua franca of post-Apartheid South Africa, and its acceptance and use
have become symptomatic of new national ideologies of which the notion of positive
discrimination in favour of non-Whites is a defining tenet. The question which we wish
to ask here is twofold: First, in how far is it possible to distinguish a Coloured from a
White norm of code-switching? Second, how far can these norms of code-switching
provide a background against which White and Coloured identities can be negotiated?
This article is organized as follows: First, we will briefly present the divisions in
the Afrikaans speech community in ethnic and linguistic terms. Then we will provide an
overview of the treatments given to identity in speech variation and their relevance to
the task of identifying contrasts relevant to (ethnic) identity in patterns of codeswitching. We will then attempt to identify White and Coloured norms of codeswitching at the grammatical and sequential levels, against the background of which we
will identify the uses individual speakers make of code-switching as a stylistic resource
to negotiate their identities.
1. Ethnic identity and language in the Afrikaans speech community
Afrikaans has its origins in the 17th and 18th centuries Cape linguistic continuum,
ranging from acrolectal to basilectal varieties of Dutch. White settlers, Slaves and Free
Blacks of various origins, as well as Khoi-Sans, contributed through their mutual
interaction to shaping the primary characteristics of Afrikaans, which were later
consolidated and generalized through a standardization scheme led by Whites from the
late 19th century onwards (Roberge 2004). The elements of a linguistic divide running
along ethnic lines were present from the inception of the Cape Dutch speech
community: Knowledge of the acrolectal varieties was the preserve of White society,
who had, more than non-Whites, access to Metropolitan Dutch, whereas native or nonnative basilectal versions of Dutch, Low Portuguese and Malay were deeply ingrained
in non-White society (Ponelis 1993: 15-17).
The British takeover in 1806 brought a number of changes to Cape society, two
of which are particularly relevant to our discussion. The first change, that took some
time to take effect, was the arbitrary creation, originally for colonial administrative
purposes, of a ‘Coloured’ administrative category that over time came to subsume all
non-White Cape Dutch-speakers and acquired the trappings of an ethnic identity
(Armstrong and Worden 1979: 122; van der Ross 1979). The second change was the
introduction of English, which over time turned out to act as an agent of social
differentiation, around which not only ideological but also ethnic tensions came to
crystallize. English symbolized social advancement for all sections of Cape Dutchspeaking society regardless of ethnicity, but it came to assume a specific value for nonWhites. English was the language of liberal policies, which, among other things, led to
the abolition of slavery and the enfranchisement of a number of Coloureds at the Cape
in the 19th century. It was also, much more than Dutch, a language through which
education for non-Whites was to some extent made possible (cpr. Lanham 1978: 22-3;
White and coloured identities in Afrikaans-English code-switching
427
Malherbe 1977: 134-5). As a result, the Coloured intelligentsia soon displayed a
tendency to identify with the ideals of the British Empire and the English language,
causing the African People Organization – a political party furthering Coloured
interests in the early 20th century – to recommend its followers to ‘drop the habit as far
as possible, of expressing themselves in the barbarous Cape Dutch’ (APO 1910 in
Adhikari 1996: 8).
The advent of Apartheid and its role in strengthening the socio-economic gap
between Whites and non-Whites in general resulted in a growing ideological opposition
between Whites and Coloureds in particular. Although the Coloured population tended
to obtain favourable treatment from the ruling White-led National Party (NP) on
account of its being culturally related to the Afrikaners, parts of the Coloured
intelligentsia were more inclined to identifying with Black-dominated liberation
movements such as the African National Congress (van der Ross 1979: 76-7, 82-3).
Social and ideological differences between Whites and Coloureds also resulted in
different attitudes to Afrikaans and English. While strong support for single medium
Afrikaans education prevailed among White Afrikaans speakers, strong symbolic
attachment to English remained conspicuous in sections of the Coloured population
despite the policy of ‘Afrikanerization’ through education that it was subjected to by the
NP (Malherbe 1977: 139; Steyn 1980: 265). Evidence of different language ideologies
across the ethnic divide was found in the general disapproval of English linguistic
influence among the White Afrikaans population (Dirven 1992: 39-40), while
increasing identification with Afrikaans-English code-switching was observable among
Coloureds, most especially in the Cape Town region and among sympathizers of the
liberation movements (Scheffer 1983: 97; Esterhuyse 1986: 117; McCormick 2002: 99100).
Nowadays, Coloureds form the majority of Afrikaans speakers with 54% of the
totals while the share of Whites is 42%. There is still a wide economic gap between
White and Coloured Afrikaans speakers. By 1996, the former had managed to preserve
their middle-class position with almost half of their effectives in semi-professional or
high occupational categories. In terms of education, almost one third of White Afrikaans
speakers were university or college graduates (Giliomee 2004: 627). By contrast,
Coloureds were in 2001 on the average still earning little more than one fifth of the
average annual income of White Afrikaans speakers. Only 15.8% fell in semiprofessional or high professional categories. Only 4.9% possessed tertiary qualifications
(South African Statistics 2001). Despite evidence of population movements across
urban areas, the residential segregation of old has endured: In most South African
municipalities, there are White- and Coloured-dominated areas.
The democratic transition has brought with itself a new sociolinguistic context in
which English is systematically favoured at the expense of Afrikaans at the institutional
level. In the event, the visibility of Afrikaans has steadily decreased as a language of
public life, business, education and media (Giliomee 2003). Increasing exposure to
English, and also ideological stigmatization of Standard Afrikaans, seem to have led to
the valorization of Afrikaans-English code-switching among upcoming Afrikaansspeaking generations (Webb 2002: 30). This observation suggests that code-switching is
spreading across the ethnic divide. It does not clarify, however, whether code-switching
is found in the same forms and functions across the ethnic divide.
428
Gerald Stell
2. Code-switching and identity
Social identity and its place in linguistic variation has formed a bone of contention
opposing theoretical outlooks derived from the variationist tradition on the one side and
approaches influenced by social constructivism on the other. Whereas the former has
been regarding linguistic variation as a predictable reflection of macro-social structures
while downplaying or completely neglecting the identity dimension, the latter has
placed speaker agency centrally, regarding linguistic variation as partly resulting from
strategic choices made for the purpose of constructing or negotiating social identities at
the level of the interaction (cpr. Coupland 2002: 188; Eckert & Rickford 2002: 3-5;
Gumperz & Gumperz 2007: 481). Both types of approaches have been taken with
regard to the relationship between ethnicity and linguistic variation, as illustrated among
other things by the large corpus of research performed on African American Vernacular
English (AAVE). Variationist treatments have tended to reduce ‘Blackness’ in the U.S.
context to a mere parameter of variation, which, in specific combinations with the social
class and context variables, is taken to predict frequencies of AAVE features. ‘Black
style’, as ‘styles’ in general, is then often presented as a quantitative and collective
measure determined solely by factors external to the speaker1. By contrast, social
constructivist approaches to Black ethnicity such as that taken by Ervin-Tripp (2002)
have emphasized the creative use occasionally made of AAVE features for rhetorical
purposes, irrespective of social class or context. From this point of view, ‘Black style’,
as ‘styles’ in general, is a resource exploited by the speaker to achieve specific
communicative ends. Social constructivist approaches to ethnicity in linguistic variation
have also been taken in situations involving code-switching between different
languages, i.e. the type of situation that we are concerned with in this present study,
such as in particular Bailey (2007) on Spanish-English code-switching among
Dominicans in the U.S.
Despite its empirical justification, the principle of speaker agency in linguistic
variation cannot be considered to operate in complete autonomy. In criticism of
Coupland’s speaker-centred approach to style (2002), Rickford (2002: 223) takes a
defensive stance on aspects of the variationist heritage, as in particular its collective
view of style, warning that ‘group styles and individual styles are both realities, and
each can help us understand the other’. The implication here is that language practices
of individuals must be socially interpreted against the background of social conventions
surrounding them.
There are indications that code-switching, as a dimension of language variation,
should also be looked at from both a collective and an individual perspective in order to
determine its relevance to the negotiation of individual identities. A practical illustration
is provided by ‘mixed codes’, i.e. conventionalised varieties characterized by high
frequencies of code-switches which are not perceived as socially meaningful, or
‘salient’ by the speakers (cpr. Auer 1998: 20; Alvarez-Caccamo 1998: 38-40). If ‘mixed
codes’ are found to be group norms, as it is the case in many African communities (cf.
1
An example of variationist treatment given to variation between Black and White vernaculars
is found in Labov (1987). Trends of divergence between the two are accounted for as ‘an effect of
residential, economic and educational segregation’ (22). Other factors, such as identity-related factors,
hardly come into play, even though Labov at one point alludes to the possibility that ethnic differences in
speech may be consciously cultivated, observing that ‘the local [Philadelphia] vowel system is an
important claim to local rights and privileges, which blacks are not prepared to make’ (20).
White and coloured identities in Afrikaans-English code-switching
429
Meeuwis & Blommaert 1998), then the potential for individual instances of switching
between languages A and B to form strategic choices relevant to identity-negotiation is
diminished. This brings us to the question of how to distinguish in practice between
those switches potentially relevant to identity-negotiation and other switches.
A point of departure for determining a group style is a quantitative survey of
grammatical code-switching types. An obvious reason for this is that the only typologies
of code-switching that have been quantitatively implemented are grammatical. A
grammatical typology of code-switching that can be quantitatively implemented while
offering the possibility of guessing at levels of social salience is Muysken’s threefold
distinction between insertions, alternations and congruent lexicalization (1997).
Quantitatively implementing Muysken’s typology on bilingual data can produce scales
of grammatical pattern dominance which can reveal whether a given grammatical type
of code-switching is more dominant than another within a given group or across groups
(cf. Deuchar et al. 2007). Following a gestalt-psychological logic, there is a distinct
possibility that salience and social meanings may be identifiable primarily in those less
dominant grammatical types (cpr. Auer 1998: 20). To give a practical illustration,
clausal alternations (a subtype of alternations in Muysken’s typology) may possess
salience, and therefore social meaning, in bilingual speech for which congruent
lexicalization (a grammatical type of code-switching potentially indicative of a ‘mixed
code’2) is a dominant pattern.
Sequential types of code-switching form another quantifiable aspect of codeswitching by which group norms may be illustrated and cross-compared. It may, for
example, be speculated that the sequential type of code-switching that Auer (1998: 8)
calls ‘preference-related code-switching’, involving the resetting of the language of
interaction in the course of an exchange, is more typical of bilingual speech
communities with no tradition of language compartmentalization. Directly related to the
sequential domain of analysis, and also quantifiable, are the contextualisation functions
assumed by language A and language B, which may not conform to the same patterns
across bilingual speech communities: Language B may specialize in adding emphasis to
the preceding information in language A in one given community while it may
specialize in downgrading the preceding information in language A in another
community (Sebba & Wootton 1998: 275).
Identities may be partly reconstructed on the basis of contrasts between surveys
of the distribution of grammatical and sequential patterns across samples. But identity
may also be reconstructed on the basis of individual instances of code-switching, which
form part of individual strategies of ‘opposition-building’ for the purpose of identitynegotiation. Individual switches may among other things serve the enactment of specific
personae, alternatingly embodying values linked to the in-group or to the out-group
(cpr. Gumperz 1982: 66; Coupland 2007: 221). Yet, individual switches may not all be
relevant to the negotiation of identities: Identity-relevant social meanings may instead
reside in ‘constellations’ of features ‘which are interpreted together’ (Auer 2007: 12).
This means that, in order to ascertain the identity-relevance of individual instances of
code-switching, a holistic point of view needs to be taken, by which the relationship
between an individual switch and its wider conversational and social environment must
be taken into account. A question to be asked here is how far individual strategies of
2
Muysken does not use the term ‘mixed code’ in his typology, yet he considers the District 6
Coloured data that McCormick (2002) refers to as a ‘mixed code’ as indicative of ‘congruent
lexicalization’ (2000: 148).
430
Gerald Stell
identity-negotiation can be implemented through code-switching in a context where
code-switching is the norm and may therefore lack stylistic salience.
We intend to reconstruct White and Coloured identities in code-switching using
a two-step approach. First, we want to determine how far ‘Coloured’ and ‘White’
identities are indexed by distinct code-switching norms. In the process, we take
grammatical and sequential patterns of code-switching as a point of departure, which we
treat as essential in the identification of potential identity-relevance in code-switching.
Second, we contrast the obtaining picture of collective ‘Coloured’ and ‘White’ codeswitching norms with a constructivist representation of code-switching strategies
meaningfully employed by individual speakers to negotiate their ethnic identities in
South Africa’s current post-Apartheid context.
3. The corpus and its sociolinguistic context
Reflecting our two-step approach aimed at determining collective code-switching norms
first and then reconstructing on the basis of individual instances how code-switching is
used creatively as a stylistic resource for ethnic identity-negotiation, our corpus is
organized in two parts. The first part is large enough to allow quantitative analysis,
while the second part consists of select individual conversations.
Two samples comprising roughly 150,000 words and 174 speakers were
collected from the Cape Flats and Gauteng respectively. The data from the Cape Flats
form our ‘Coloured’ sample, whereas the data from Gauteng form our ‘White sample’.
Underlying our categorization into ‘White’ and ‘Coloured’ is the informants’ selfascription of these labels. There are several reasons why we consider these two samples
comparable even though they do not originate in geographically contiguous regions.
First, Cape Flats Coloureds and Gauteng Afrikaans-speaking Whites both share a long
tradition of Afrikaans-English bilingualism. Second, the younger generations of these
samples mostly consist of students attending academic institutions which are officially
English-speaking (e.g. the University of the Western Cape), or in the process of
becoming de facto English-speaking (e.g. the University of Pretoria or the University of
Johannesburg).
Each sample is comprised of roughly equal numbers of males and females and
can be broken down in three age cohorts (17-25, 25-50, 50+). The recruitment of
informants mostly followed the ‘friend of a friend’ approach (Milroy 1987). Since the
guideline for data collection was to minimize the Observer’s Paradox in order to obtain
a representative picture of spontaneous speech in intra-group communication, the task
of interviewing was left to insiders sharing the ethnic background and age range of the
informants. All younger speakers were recruited from the university campuses of the
University of the Western Cape (UWC), the University of Pretoria (UP) and the
University of Johannesburg (UJ). In terms of social categories, the older and median age
cohorts comprised in the samples can be characterized as representative of average
income levels within the White and the Coloured communities respectively3.
3
Average income levels are documented in reference to ethnic groups by South African
Statistics (2001). The reason for using White and Coloured average income levels as a point of departure
seems more appropriate than general indicators which may lead to comparisons between social categories
not symmetrically distributed across the ethnic divide.
White and coloured identities in Afrikaans-English code-switching
431
All younger members of our White sample enjoyed comprehensive Afrikaans
medium instruction. Some members of the Cape Flats Coloured sample enjoyed
comprehensive Afrikaans-medium education, others followed an all-English
curriculum4.
In order to carry out a stylistic analysis with special focus on individual
motivational processes relevant to the negotiation of Coloured and White identities, we
singled out a number of conversations in which the activity of ethnic self-categorisation
is most inferable. These conversations involve White students from UP and UJ
respectively, and Coloured students from UWC. The topics around which these
conversations are organized may overtly involve the meaning of being White or
Coloured in the post-Apartheid context, but they also involve other topics which we
regard as more or less directly relevant to the definition of White and Coloured identity,
as in particular religious life and relations to the ‘outgroup’ embodied in abstract entities
such as among other things policies of Affirmative Action or university administration.
The question we want to answer is whether, in reflection of possible differences in
grammatical and sequential patterns of code-switching, the use of English has different
implications for the negotiation of ethnic identities across the White/Coloured divide.
4. Grammatical patterns of code-switching
The data proved to be lexically dominated by Afrikaans (over 97% of the lexicon), but
the proportion of English items in the Coloured data is higher than in the White data,
and also higher among younger than older speakers (roughly 9.5% among the younger
speakers of our Cape Flats sample versus 3.37% in the Gauteng White sample). If we
leave discourse markers out of consideration, it seems that the corpus is characterized
by unidirectional switching (Afrikaans towards English) with a few exceptions (English
towards Afrikaans) found in the Cape Flats sample. This present section is largely based
on Stell (2009), to which I refer for methodological aspects such as the identification of
switches. It first deals with insertions (4.1), followed by alternations (4.2) and congruent
lexicalization (4.3).
4.1. Insertions
In the insertion pattern, one language A will usually dominate by determining the
overall structure into which constituents from language B are inserted (cf. example 1
from our corpus where the English insertion is in bold, see further Muysken 1997: 361).
(1) Vir my is dit net betjie boring.5
‘For me it is just a bit boring.’ (Gauteng sample)
Insertions clearly form the dominant code-switching type in the data in general,
yet they occur more frequently among Coloureds than among Whites. Also, they occur
more frequently among younger than older speakers, although frequencies are
4
No distinction was made between the two groups since they seemed to belong to the same
social networks.
5
The switches under discussion are in bold, except when they are not relevant to the discussion.
432
Gerald Stell
comparably high across all three generations of the Cape Flats sample. A number of
contrasts between samples can be observed with respect to the nature and form of
insertions. One among these contrasts concerns the grammatical nature of insertions:
English adverbs are represented primarily in the Cape Flats sample.
Another area of contrast concerns the levels of morphosyntactic integration of
insertions. There is, for example, a tendency in the Cape Flats Coloured sample for
using English adjectives more frequently in attribute position where they may require an
inflection (which in fact they rarely receive, unlike in example 2) than in the other
samples.
(2) Dit is die stupidste ding wat ek in my lewe van gehoor het.
‘This is the most stupid thing I have heard of in my life.’
(Cape Flats sample)
Also typical of the Cape Flats Coloured sample is the occasional use of English
past participial affixes in predicate position, or the absence of affixes, or the use of an
English past participial affix concurrently with an Afrikaans affix (4), while in the
Gauteng White sample the Afrikaans affix dominates in the same context (3).
(3) Nee hy is nie ge-drop nie hy het nie side-fenders nie hy is heel standaard.
‘No it is not dropped it has no side-fenders it is all standard.’
(Gauteng sample)
(4) Die persoon wat my ge-registered het, het nie vir my gehelp nie.
‘The person who registered me didn’t help me’. (Cape Flats sample)
Flagging of insertions in the sense of insertions signalled by metalinguistic
commentaries, is illustrated by (5), and occurs mostly among middle-aged and older
Gauteng White speakers.
(5) Nou my vrou het ’n gout, verskoon die Engels.
‘Now my wife has a gout, excuse the English.’ (Gauteng sample)
4.2. Alternations
In the alternation pattern, both languages A and B occur alternately, each with their own
structure, with the switch point being located at a major syntactic boundary (ex. 6, see
further Muysken 1997: 361). The implication is that there may be two matrix languages
within one single utterance as long as the transition between the two is signalled by a
salient syntactic boundary. Alternations may be found in the form of a discourse marker
(example 7 from our corpus) or of a fully-fledged clause more or less loosely connected
to the preceding one (example 6 from our corpus).
(6) Toe discuss ek met my vrou en toe los ek my werk en ek apply hier by die
university en the rest as they say ja is history.
White and coloured identities in Afrikaans-English code-switching
433
‘Then I have a discussion with my wife and then I resigned from my job and I
applied here at the university and the rest as they say ja is history.’
(Cape Flats sample)
(7) Oh well dis wat gebeur as jy deel is van die Paaikikerikie Express.
‘Oh well this is what happens if you are part of the Paaikikerikie Express.’
(Cape Flats sample)
Alternations in the form of discourse markers6 are more or less evenly
distributed across samples, and also to a lesser degree across generations. Clausal
alternations, on the other hand, are distinctly more represented in the Coloured sample,
and among younger speakers in general. Clausal alternations are most marked
generationally in the Gauteng White sample, and least marked generationally in the
Cape Flats Coloured sample. Switches involving clausal alternations are less frequent,
and therefore presumably more salient and likely to carry a social function than
insertions.
A prominent area of contrast across samples with respect to alternations is found
in the nature of the syntactic boundaries at which they occur. Alternations occurring at
coordination or subordination sites (6) are roughly evenly distributed across samples.
On the other hand, English main clauses followed by Afrikaans subordinate clauses (8)
are only present in the Coloured sample, and switched clauses with no formal relation to
the neighbouring clauses7 are proportionally by far better represented in the White
sample (9).
(8) So let us say jy bly in ʼn plek soos Manenburg dan kan jy nie jy kan jy kan
nie ʼn loan by die bank kry nie.
‘So let us say you stay in a place like Manenburg then you can’t you can’t
you can’t get a loan from the bank.’ (Cape Flats sample)
(9) Maar dis jou eie besluit of jy dit sal toelaat maar ek weet nie ek het geen
probleem met enige swartmense bruinmense pienkmense nie. Live and let live.
‘But it’s your own decision whether or not you will allow that but I don’t
know I have no problem with any black people brown people pink people. Live
and let live.’ (Gauteng sample)
4.3. Congruent lexicalization
With congruent lexicalization, ‘the grammatical structure is shared by languages A and
B, and words from both languages A and B are inserted more or less randomly’
6
Many English discourse markers form established borrowings in Afrikaans. All English
discourse markers listed in the AWS 2002 under the heading ‘Colloquial Afrikaans’ were left out of
consideration here.
7
Muysken suggests that the switch point of alternations is necessarily located within the
‘clause’, while elsewhere he describes alternations as possibly occurring between ‘utterances’ (cpr. 2000:
33, 96). It has been chosen here to adopt the broadest possible definition of alternation and to include in
the concept those utterances in language B which occur within turns and are formally unrelated to the
neighbouring utterances in language A by means of coordination or subordination.
434
Gerald Stell
(Muysken 2000: 8). Non-constituency is a defining feature of congruent lexicalization,
as illustrated by example 10 from our corpus, but it is not the only one: There is also
‘back and forth switching’, which may refer to ‘rapid succession of switches’ in the
form of insertions and/or alternations (cpr. Muysken 2000: 132-3, 237).
(10) Ek suck at dit.
‘I suck at it.’ (Cape Flats sample)
Two forms of non-constituents were met in the data, i.e. clause-internal nonconstituents as in (10) and clause-peripheral non-constituents with an alternational
character in the form of clause-final elements + discourse markers, as illustrated by
(11).
(11) Jy is ’n daily obviously.
‘You are a daily obviously.’ (Gauteng sample)
The Coloured sample is the only sample with both types, while the White
sample only has the type illustrated by (11).
With ‘back and forth switching’, the likelihood of congruent lexicalization can
be identified only at corpus-level rather than at the level of the utterance. The ‘global’
diagnostic features of congruent lexicalization which are most relevant to our data are 1)
there is a high grammatical diversity of switches, 2) bidirectionality of switching is
high, 3) non-constituents are common (Muysken 2000: 230). As mentioned earlier, 2) is
hardly found in our data, but when it is, it is in the Cape Flats Coloured sample, which
also musters the widest grammatical array of switches and the highest frequency of nonconstituents. For these reasons, it is the Cape Flats Coloured sample that displays the
highest potential for ‘congruent lexicalization’, and therefore for a ‘mixed code’ where
all switches, including both insertions and alternations, are likely to lack salience and
social meaning. By implication, salient and socially meaningful code-switches are more
likely to be found among Whites, most especially in the form of clausal alternations.
5. Sequential patterns of code-switching
If speaker turns consisting only of an English discourse marker are left out of
consideration, Afrikaans, or a ‘mixed code’ with Afrikaans as an identifiable matrix, is
for all intents and purposes the only language-of-interaction with no sustained
divergence of language choices. The dominance of Afrikaans across turns implies that
Afrikaans-English code-switching in our data is mostly turn-internal, which according
to Auer (1998: 17) forms one of the characteristics of a ‘mixed code’. This need not
mean, however, that turn-internal code-switches in our data lack the salience to fill
contextualising functions. In fact, contextualisation functions are generally inferable
from a large number of code-switches, most especially clausal alternations. In this
section we are looking at turn-internal clausal alternations in turn-peripheral and midturn positions (5.1 and 5.2), as well as between-turn code-switching (5.3).
White and coloured identities in Afrikaans-English code-switching
435
5.1. Code-switching in turn-peripheral position
Among all cases of alternations, the most frequent type is turn-peripheral in the form of
either turn-initial or turn-final English discourse markers or clausal alternations (cf. 4.2).
Table 1 shows that the turn-initial and turn-final positions for English alternations
(excluding quotations) is almost identical across the two samples in that the turn-final
position seems to dominate in both cases.
Table 1 Turn-peripheral English alternations per sample (younger age cohorts, %)
Position
Gauteng White
Cape Flats Coloured.
37%
37.5%
Initial
63%
62.5%
Final
A striking contrast between the White and the Coloured sample is concerned with the
proportion of discourse markers in initial and final English alternations: The share of
discourse markers is proportionally larger in the White sample than in the Coloured
sample, while the share of clausal alternations is conversely higher in the Coloured
sample than in the White sample.
Table 2 Proportions of discourse markers and clausal alternations in total number of
turn-peripheral alternations (younger age cohorts, %)
Gauteng White
Cape Flats Coloured.
77.4%
28.5%
DM
22.6%
71.5%
Clausal
Generally, clausal alternations are more likely to be found in turn- or utterance-final
position (cf. example 9) than in turn- or utterance-initial position (cf. example 8). The
only sample where clausal alternations may be found in turn- or utterance-initial
position is the Coloured sample.
Another area of contrast concerns the levels of syntactic embeddedness of turnor utterance-final clausal alternations. In all cases found in the White sample, they are
not syntactically connected to the preceding clause, and are therefore of a paratactical
nature. Example 9 illustrates this point, with the English clause live and let live being
connected to the preceding Afrikaans clause only by forming part of the same
intonational unit8. By contrast, example (6) illustrates the higher level of syntactic
embeddedness found with turn- or utterance-final clausal alternations in the Coloured
sample, where overt syntactic markers used as coordination devices in particular
connect the English clause to the preceding Afrikaans clause. There are also cases
where a turn- or utterance-final English clausal unit is syntactically interwoven with the
preceding Afrikaans clause by being ‘triggered’ by an English element comprised in the
latter, as shown in example (12) where prepositionally used like introduces an English
clause which then introduces yet another clause. Such instances of ‘triggering’
constitute defining characteristics of congruent lexicalization or of a ‘mixed code’ (cpr.
Muysken 2000: 231, Auer 1998: 17).
8
Cf. Hopper & Closs Traugott (2003: 180-1) for intonational cues as means to identify parataxis.
436
Gerald Stell
(12) So en mens hum is amper soos (.) ek weet nie dis like spitting in the wind
when you are dealing with the administration at this university.
‘So and one is erm almost like (.) I don’t know it’s like spitting in the wind
when you are dealing with the administration at this university.’
(Cape Flats sample)
Contrasts in degrees of syntactic embeddedness seem reflected in contrasts in
pragmatic embeddedness. Turn- or utterance-final clausal alternations are often linked
in a metonymic relationship to the preceding clause in the White sample while in all
cases forming ‘closing sequences’ (cf. Alfonzetti 1998: 188-9). This is again illustrated
by example (6) where the English clause live and let live is a catchphrase used by the
speaker to subsume her previous argument while also signalling the end of her turn. By
contrast, the English chunk like spitting in the wind when you are dealing with the
administration at this university in example (12) could possibly be read as a metonymy
for preceding observations (not reproduced here) only if taken together with the
Afrikaans elements it is introduced by. In other words, it is the whole bilingual
sequence in (12) that could be read as a metonymy (as well as possibly a ‘closing
sequence’), not just the English chunk, which does not possess enough local syntactic
and pragmatic salience.
5.2. Code-switching in mid-turn position
The share of alternations occurring in mid-turn position is proportionally larger in the
Coloured sample than in the White sample (38% versus 70% in the Coloured sample,
excluding quotations). The imbalance between the two samples in this respect is also
reflected, to a lesser degree, in the distribution of alternations occurring within
utterances (23% in the White sample versus 37% in the Coloured sample, excluding
quotations). A more striking contrast resides in the nature of mid-turn alternations,
which happen to much more frequently consist of clausal alternations in the Coloured
sample (cf. example 14) than in the White sample, where discourse markers are more
dominant (cf. example 13).
(13) Ja, want baie mense gaan werk actually by plekke wat wat hulle glad nie geswot
het.
‘Yes because lots of people actually take up employment for which they have not
studied.’ (Gauteng sample)
(14) Hulle het ʼn policy gehet van van redlining so basically wat daai beteken is if
you live on the other side of there you know arbitrary demarcated red line dan
kan jy nou nie ʼn homeloan kry nie irrespective of your credit history.
‘They had a policy of of redlining so basically what that means is if you live on the
other side of there you know arbitrary demarcated red line then you can’t get a
homeloan irrespective of your credit history.’ (Cape Flats sample)
White and coloured identities in Afrikaans-English code-switching
437
As already mentioned in 5.1, there is a possible tendency of ‘triggering’ in the Coloured
sample, whereby the use of an English element causes the following element to be in
English as well. This is among other things illustrated by the use of English discourse
markers, which tend to be combined in the Coloured sample as in so basically9 (with so
phonetically realized along the English pattern10) in (14) above. That tendency of
triggering and the relative lack of syntactic salience that it entails for English clausal
alternations seems to again result in a relative lack of pragmatic salience. In other
words, if we leave aside quotations (cf. 5.4), most clausal alternations in the Coloured
sample seem not to fill any contextualisation function except when they form selfrepairs. By contrast, the few mid-turn clausal alternations (excluding quotations) that we
find in the White sample can be sequentially accounted for as either ‘pre-closings’ (cf.
Alfonzetti 1998: 188-9) or self-repairs.
5.3. Between-turn switches
What we refer to as ‘between-turn switching’ here is single English turns with which
our data are occasionally interspersed. Most instances of English between-turn
switching are formed by English discourse markers such as ok, allright/oraait, cool,
etc., which form acknowledged borrowings in Afrikaans (cf. 4.2). In the remaining
cases, we find English clauses ranging from nominal to complex. If we consider
between-turn switches to form special cases of alternation, then their share in the total
number of alternations is proportionally higher among Whites than among Coloureds
(23% versus 7%). An explanation for this could be that the tendency among Coloureds
to use an Afrikaans-English ‘mixed code’ (cf. 5.1) causes all-English between-turn
switches to be less likely to occur than among Whites. Evidence for the occurrence of
all-English between-turn switches being hampered by that tendency to use a ‘mixed’
Afrikaans-English code is shown in example (15) where B’s response to A’s selfjustification about HIV testing starts in English in what seems to be a beginning
‘closing sequence’ (cf. 5.1) but then carries on into Afrikaans past the dual language
form is which forms the triggering factor here11.
(15)
A
As ons nou saam is gaan toe sê hy vir my hy gaan elke keer Botswana toe en dan gaan
hy dan gaan hy dalk uit wees maar hy sê nou nie hy gaan op my cheat nie but ek moet
versigtig wees en ek moet (.) ek moet aanhou om elke drie maande te gaan vir HIV
testing no no mama...
B
9
Note that, since it does not form a constituent, the English stretch redlining so basically, could
count as an instance of congruent lexicalization (cf. 4.3). Yet, that instance of code-switching may
alternatively be read as an insertion (in the form of the substantive redlining) followed by an alternation
(in the form of the combined discourse markers so and basically).
10
The English connective so has an Afrikaans homograph realized as [suə] which may be used
in comparable functions in colloquial Afrikaans speech, even though its generalized usage along English
lines is not recognized by Afrikaans lexicographic sources.
11
The fact that the use of English in that example is not co-extensive with the closing sequence
is reverberated elsewhere in the use of English to introduce the first part of a quotation rather than to
report the whole quotation.
438
Gerald Stell
I think I think the moral of the story is jy moet jou (.) dit is jou eie responsibility
always as (.) individuals het ’n personal responsibility, aanhou jouself om maar jouself
te protect.
A
When we are together he tells me he goes every time to Botswana and then he may go
out but then he doesn’t tell me he is going to cheat on me but I must be careful and I
have to keep on going for HIV tests every three months no no mummy...
B
I think I think the moral of the story is you must (.) this is your responsibility always as
(.) individuals have a personal responsibility, carry on protecting yourself.
(Cape Flats sample)
Between turn switches involving clauses differ across the two samples in terms of their
sequential position. While they may occur in both initiative and responsive turns in the
Coloured data, between turn switches involving clauses only occur in responsive turns
in the White data (16). Another contrast between Coloured and White usage worth
mentioning has to do with the semantic contents of between-turn switches and their
interactional function. Between-turn switches involving clauses in the White sample
often consist of aphorisms used as ‘back-channels’ in a metonymic relationship with the
previous speaker’s utterance, as illustrated by (16).
(16)
A
Maar hoor net gou. Weet jy? Weet jy hoekom sê sy sy is slim? Want sy het hum net
deur daardie single sin wat sy gesê het dat mense die show in die eerste plek kyk om
actually te sien hoe dom sy is.
B
There is no such thing as bad publicity.
A
Presies.
A
But listen. Do you know? Do you know why she says she is clever? Because she erm
only by that single sentence she said that people watch the show in the first place to
actually see how stupid she is.
B
There is no such thing as bad publicity.
A
Precisely. (Gauteng sample)
In this example, B’s reaction to A assumes the form of a catchphrase acting as
metonymic acknowledgement of the information provided by A. B’s switching to
English does not imply the resetting of the language-of-interaction, as the exchange
carries on in Afrikaans from A’s Afrikaans back-channel onwards.
White and coloured identities in Afrikaans-English code-switching
439
6. The construction of identities through code-switching
The preceding sections have focused on grammatical and sequential contrasts between
Coloured and White code-switching practices. A range of contrasts have been identified
at group-level, such that we may claim to have a representation of ‘Cape Flats
Coloured’ and ‘Gauteng White’ code-switching styles. Now the question arises how
code-switching is meaningfully used by individual speakers to negotiate their identities.
This section examines how individual uses of code-switching vary according to younger
speakers’ self-positioning in relation to topics which are relevant to the representation of
White and Coloured identities in the current South African context. Against the
background of these topics it first attempts to reconstruct the identity-relevance of codeswitching involving opposition between monolingual English code and Afrikaans or
mixed Afrikaans-English codes (6.1), and then focuses on the identity-relevance of
‘mixed’ Afrikaans-English codes (6.2).
6.1. Identity and English monolingual code
Younger generations of Afrikaans-speakers currently find themselves in a challenging
sociocultural context where the notion of ‘South African-ness’ is being redefined to a
certain degree at the expense of the White and Coloured ethnic groups through among
other things Affirmative Action policies based on the racial categories upheld under the
Apartheid. Positive discrimination in South Africa inevitably implies the maintenance
of White and Coloured ethnic stereotypes with which identification is problematic since
it may be interpreted as the rejection of ‘New South African’ values. We find among
both White and Coloured speakers instances clearly illustrating the association made
between English monolingual code and ‘New South African’ values. It seems that
contrasts between Afrikaans and English monolingual clauses is a common device
among the younger generations of the Gauteng White sample for claiming parts of a
‘White’ heritage while endeavouring to situate it as uncontroversially as possible within
a broader South African context. A telling example is (17) below:
(17)
A
Uhm, ons praat nou daarvan, maar ek bedoel, as ons nou praat van Afrikaanse mense in
Suid-Afrika. Jy weet so, jy sou sê jy voel deel as … jy is ’n Afrikaanse mens?
B
Ja, first and full.
A
Voel jy as deel van Suid-Afrika?
B
Ja, ek is ‘n Suid-Afrikaner.
A
Erm, if we are talking about it, I mean, if we are now talking about Afrikaans people in
South Africa (.) You know, would you say you feel part of...you’re an Afrikaans person?
B
Ja, first and full
440
Gerald Stell
A
Do you feel part of South-Africa?
B
Yes, I am a South African. (Gauteng sample)
In this example, A is quizzing B about his stance towards the Afrikaans heritage, which
in this conversation is mainly expressed in terms of language loyalty. Being ‘Afrikaans’
in Gauteng does however inevitably have a strong White connotation (since Whites
locally form the overwhelming majority of Afrikaans speakers). B’s English response to
A’s question must be read against the background of that obvious fact as well as B’s
subsequent claim to South African-ness. (White) Afrikaans and South African identities
tend to be perceived as being mutually exclusive in the current context, as suggested by
Roefs (2006: 81-2), and it is in order to lift that perceived incompatibility that B
rhetorically lays claim to his (White) Afrikaans heritage by means of English, i.e. the
language associated with South Africa’s public sphere and ANC discourse (Orner 2008:
99-100). In other words, what speaker B does is signalling affiliation with Afrikaans
and Whiteness while tempering the ideological implications of that affiliation by fitting
it in a ‘New South African’ frame of interpretation subsumed in the use of English.
Monolingual English code in the form of clausal alternations or between-turn switching
can be found in similar functions among younger Coloured speakers, who may exploit
the contrast it offers with monolingual Afrikaans or mixed Afrikaans-English codes. An
illustration is found in example (18) where the speaker summarizes his point of view
with regard to the discrimination faced by Coloureds in the implementation of the
Affirmative Action policy.
(18)
1. Almal wil nou die equity numbers regkry. Dit gaan alles oor nommers. Mense
2. moet nou hulle nommers regkry en die deadline kom, so basically ek hou nie
3. van die terms nie maar mense moet maar seker descriptive wees en nou moet
4. jy net...dis nou net Africans of black people wat nou inkom verstaan jy en daai
5. create daai create nogals ʼn groot fiction maar ek dink Affirmative Action (.) Ja
6. mens moet seker maar re-define what is black and what is white nè if you
7. are not white you are black (.) ja because daai is maar bietjie van ʼn probleem.
1. Everybody wants to get their equity numbers right. It’s all about numbers.
People
2. have to get their numbers right and the deadline is there, so basically I don’t
like
3. the terms but people surely have to be descriptive and now
4. you just have to...it’s only Africans or Black people coming in do you
understand and that
5. creates that creates quite a big fiction but I think Affirmative Action (.) Yes
6. one surely has to re-define what is black and what is white right if you
7. are not white you are black (.) Yes because that is a bit of an issue.
(Cape Flats sample)
Following a description of the Affirmative Action policy (line 1 to 2) the speaker states
his opposition to it. He tempers his opposition with a concessive move which he
White and coloured identities in Afrikaans-English code-switching
441
interrupts with the racially restrictive observation bordering on indictment that it is only
black people who benefit from the policy. However, the thrust of the whole argument is
not a wholesale condemnation of Affirmative Action. Rather, the speaker calls for a
redefinition of the policy along new terms equally favourable to both Blacks and
Coloureds. The use of English across 6 and 7 coincides with that call for a redefinition
of Affirmative Action’s ethnic nomenclature, to which it adds a dimension of solidarity
with the current beneficiaries of that policy, i.e. ‘Africans’ or ‘black people’.
‘Colouredness’ is thus projected in opposition to ‘Whiteness’ and in affiliation with
‘Blackness’. Just as in (17), monolingual English code is projecting an identity-relevant
stance through an interpretive frame in which solidarity with a racially blind or Black
South African-ness is emphasized.
In summary, English monolingual code may in both samples imply some degree
of identification with ‘New South African’ values. But there remains a major contrast
between samples in the identity-relevant functions assigned to ‘mixed codes’.
6.2. ‘Mixed codes’ and identity
We established in 4.3 that there are symptoms of congruent lexicalization in both the
Gauteng White and the Cape Flats Coloured samples. These symptoms suggest the
possible use of a ‘mixed’ Afrikaans-English code, which may meaningfully contrast
with a monolingual English code. But we do not yet know how far there may be a
question of a ‘mixed’ Afrikaans-English code meaningfully contrasting with a more
monolingual Afrikaans code. If any such alternation takes place, could it be used along
the lines predicted by Gumperz (1982: 66) to project an opposition between a ‘wecode’, associated with more monolingual varieties of Afrikaans, and a ‘they-code’,
associated with Afrikaans varieties with a stronger English input?
Many bilingual communities have been reported to recurrently use ‘mixed
codes’ as ‘recurrent cues’ to given episodes or invoked social identities. But the
composition of such ‘mixed codes’ may fluctuate in reflection of changing situations
and communicative goals (cf. Alvarez-Caccamo 1998: 39). This is reflected in our data
by evidence that (mostly White) speakers do to different degrees operate stylistic
choices by modulating the English input in their Afrikaans within the same situational
contexts, resulting in individual ranges of distinct styles, of which the contrastive value
may serve the negotiation of identity.
The language use of one female speaker (whom we will refer to as ‘A.R.’), who
is referred to as ‘non-conservative’ by the other participants, provides a revealing
illustration, in that it displays different levels of English input across ideational focuses
and personal stances. When quizzed about her feeling versus Afrikaans as a language,
A.R. takes a clear affiliative stance illustrated by comments such as ‘I prefer to write in
Afrikaans’, ‘there are more descriptive terms in Afrikaans than there are in English’. In
reflection of much of the surrounding speech material, this affiliative stance is conveyed
through a monolingual Afrikaans code. A stylistic contrast is created when, quizzed
about her feelings versus certain Afrikaans stereotypes, A.R. takes a clear disaffiliative
stance, linguistically signalled by English input (19).
(19) Dit is regtig...jy kry...dis net jou hardcore uhm AP Kerk Afrikaner wat sou
Kyknet kyk reg deur die dag oordat hulle nog vasgevang is in die voor ninety
442
Gerald Stell
four era en dis al hoekom hulle Kyknet kyk uhm want Kyknet se programme is
up to shit en daar is nie genoeg nie so jy sien hulle ses keer’n dag.
‘This is really...you get...that’s just your hardcore AP Church Afrikaner who
would watch Kyknet all day long because they are still stuck in the pre-ninety
four era and this is why they watch Kyknet erm because Kyknet’s programmes
are up to shit and there isn’t enough so you see them six times a day.’ (Gauteng
sample)
The English input here clearly has a discourse function in that it casts in a derogatory
frame a stereotypical conservative ‘Afrikaner’, i.e. the ‘hardcore A.P. Afrikaner,
characterized by radical Christian religiousness (the Afrikaans Protestantse Church
counts among the more radical Afrikaans churches). Not only for its English origins but
also for its alternative pop-rock connotations, the label hardcore collides head-on with
the A.P. stereotype, who necessarily identifies not only with Suiwer Afrikaans (‘pure
Afrikaans’) but also with past and outfashioned values. The implied conflict between
the ‘A.P. Afrikaner’ stereotype and modernity is emphasized again in the use of the
expression ninety four era, of which the English form accords with its symbolic
connotation as the end of Afrikaner dominance in South Africa. Finally, the derogatory
expression up to shit expeditively dismisses one of the defining cultural attributes of the
‘A.P. Afrikaner’, i.e. the all-Afrikaans channel Kyknet, as not only qualitatively low but
also as hopelessly out of place in a modernity dominated by Anglo-Saxon judgment
values. The constellation of English insertions here assumes the resonance of a ‘theycode’, a city sophisticate and also post-Apartheid voice, which A.R. appropriates to
distance herself from ingroup values she rejects.
There is no compelling evidence of a continuum of distinguishable AfrikaansEnglish mixed codes in the Coloured data contrastively used within the same
communicative situations. This could mean that the ‘mixed’ Afrikaans-English code
found in the data forms a ‘recurrent cue’ of Coloured identity, the use of which is not
determined by alternations between interpretive frames as exemplified by the instance
of ‘mixed’ Afrikaans-English code in (19). The assumption that the ‘mixed’ AfrikaansEnglish code is perceived as a ‘we-code’ by default, which could be one more clue that
individual switches observable on the surface are not socially meaningful in Coloured
speech, is directly suggested by those instances where it is found in contiguity with a
monolingual English clause used in an emphatic function. In example (20), the
monolingual English clause in you see this is not something that is one dimensional is
used conclusively as a ‘topper’, a function associated with the prestige language or the
‘they-code’ by many bilingual communities, suggesting that the preceding utterance –
which displays the features of a ‘mixed code’ and serves to convey pragmatically less
salient information – .possesses the typical functions of a ‘we-code’ (cf. Sebba &
Wootton 1998: 275).
(20) Dit het baie te doen met die kulturele perspective ook because daar is nog
mense wat dink as jy as jy Aids het en jy rape’n baby dit gaan jou cure you
know what I mean so sulke sulke cultural misconceptions is wat ons ook
moet... gaan nie maklik...you see this is not something that is onedimensional.
White and coloured identities in Afrikaans-English code-switching
443
‘It has lots to do with the cultural perspective also because there are still
people who think that if you have Aids and you rape a baby it’s going to cure
you know what I mean so such such cultural misconceptions are what we
also...it’s not going to be easy... you see this is not something that is onedimensional.’ (Cape Flats sample)
7. Discussion
Our study has revealed a range of similarities and contrasts in forms and functions of
Afrikaans-English code-switching across the Coloured/White divide. These can be
found both at the group-level and at the individual level. We have been able to identify
a White and a Coloured group style, against the backdrop of which certain uses made of
code-switching by individuals could be identified as relevant to the negotiation of
identity.
Code-switching forms and functions turned out to be similar to a certain extent
across the Coloured/White divide. Our survey of grammatical patterns has shown that
Afrikaans is the dominant matrix language in both samples, and that the frequency of
insertions, alternations and congruent lexicalization is increasing across generations of
both Coloured and White samples. Yet, a more detailed glance at the data reveals a
range of contrasts.
English material is morphologically and syntactically more integrated in the
Coloured data than it is in the White data. This is illustrated by the observation that
Afrikaans morphological rules are more often waived in the Coloured data, where
English adjectives occur more often in uninflected form in attribute position, and
English verbs occur more often with English past participles morphemes. By contrast,
English adjectives tend to occur in the White sample where they do not need to comply
with Afrikaans morphological rules and English verbal forms are assigned Afrikaans
past participle morphemes. Evidence of English material being syntactically more
integrated in the Coloured data is found in the higher proportion of English clauses at
subordination or coordination sites and in the more frequent mid-sentence occurrence of
English non-constituents. By contrast, English clauses are more often flagged or
detached syntactically in the White data, while non-constituents tend to be rare and
sentence-peripheral. Generally, the nature and frequencies of insertions, alternations and
non-constituents indicate that the potential for congruent lexicalization is stronger in the
Coloured data than in the White data.
The sequential characteristics of Afrikaans-English code-switching display stark
differences across the White/Coloured divide. We find in the Coloured data that
complex English input (in the form of English clauses or English non-constituents)
tends to occur more in mid-turn position, apparently not filling any contextualisation
function. When it can be interpreted as signalling a new conversational frame, the
English input is not co-extensive with that conversational frame: It may unpredictably
give way to Afrikaans or spill over into the subsequent sequential unit following a logic
of ‘triggering’. Also, we find in the Coloured data that between-turn switching tend to
occur in all sequential positions. By contrast, English clausal alternations tend more to
be turn-peripheral in the White data, and to fill contextualisation functions such as that
of closing a turn. In contrast to the Coloured data, English clausal alternations are coextensive with the new conversational frame they signal. When they occur in between
444
Gerald Stell
turns, English clauses tend more to occur in responsive turns in the White data. The
apparently freer sequential position of complex English input in the Coloured data
reflects the characteristics associated with ‘mixed codes’ in Conversation Analysis (cf.
Auer 1998: 17-20), which reflects at the conversational level the higher potential for
congruent lexicalization which we made out in the Coloured data. The more restricted
sequential position of complex English input in the White data, as well as its higher
functionality as a contextualization device is obviously reflective of another norm of
code-switching, whereby the use made of the guest language, i.e. English in our case, is
perceived as interactionally more salient.
This picture presented above summarizes the similarities and differences
between what we could term a Coloured and a White code-switching norm. These
norms provide the backdrop against which code-switching may be used by individuals
for the purpose of negotiating their identities. That use is to a large extent determined by
salience. The observation that Afrikaans-English code-switching in the Coloured data
occurs more visibly within the frame of a ‘mixed code’, of which we find symptoms
across all three age cohorts of the Coloured sample, suggests that individual switches
may lack the social meaning which is required for serving as devices for negotiating
identities. On the other hand, the syntactic detachment of English input in the White
data, as well as the observation that English input strongly increases across age cohorts
in the White sample, suggests that it is likely to serve stylistic opposition-building
among upcoming generations (cpr. Auer 2007: 14).
This prediction is met in individual examples we used. English input as used by
young White speakers was found in many instances to signal ideological disaffiliation
from traditional conservative stereotypes within the in-group context of the White
Afrikaans-speaking community. Conservative stereotypes are also present in the
Coloured community, but we found evidence of such stereotypes only in discussions
revolving around aspects of religious practices. The fact that these conservative
religious practices are not necessarily associated with Afrikaans monolingualism in the
context of the Cape Flats implies that there is less scope for English input to fill a
disaffiliative function within the in-group context in which our Coloured speakers were
observed. Rather, English input in Afrikaans in the context of the Cape Flats tends to
possess what seems to be a stylistically neutral status as observed for much of the postcolonial language input in other ‘mixed codes’ in Africa (cf. Meeuwis & Blommaert
1998). If English more visibly fills a disaffiliative function among our White speakers
than among our Coloured speakers, it is equally used to create affiliation with ‘New
South African’ ideological values, which are associated with English. We found among
other things evidence of English being used in the expression of solidarity with
multiracial ideals or with the Black majority.
Signs of convergence between White and Coloured code-switching practices are
observable at the grammatical level, and could suggest that congruent lexicalization and
a ‘mixed’ Afrikaans-English code could spread among upcoming White generations. It
is open to question, however, whether English input will lose all features of its current
opposition-building potential, as among other things its sequential specialization. In
other words, Afrikaans-English code-switching may carry on filling a variety of
identity-relevant functions more visibly among Gauteng Whites than among Cape Flats
Coloureds.
A theoretical implication of our study is that the reconstruction of the identityrelevance of code-switching is facilitated by a reconstruction of group norms.
White and coloured identities in Afrikaans-English code-switching
445
Grammatical and sequential measures for identifying ‘mixed codes’ and oppositionbuilding potentials are instrumental in making predictions as to the identity-relevant
potential of individual switches. Quantitative and qualitative approaches are thus
confirmed to be complementary in the reconstruction of identity in language variation.
References
Adhikari, M. (1996) Straatpraatjes. Language, politics and popular culture in Cape Town, 1909-1922.
Pretoria: Van Schaik.
Alvarez-Caccamo, C. (1998) From ‘switching code’ to ‘code-switching’. Towards a reconceptualisation
of communicative codes. In P. Auer (ed.), Code-switching in conversation. Language, interaction and
identity. Cambridge: Cambridge University Press, pp. 29-48.
Armstrong, J.C., and N.A. Worden (1979) The slaves, 1652-1834. In R. Elphick & H. Giliomee (eds.),
The shaping of South African society, 1652-1840. Cape Town: Maskew Miller Longman, pp. 109-162.
Auer, P. (1998) Code-switching in conversation. Language, interaction and identity. Cambridge:
Cambridge University Press.
Auer, P. (2007) Style and social identities. Alternative approaches to linguistic heterogeneity. Berlin:
Mouton De Gruyter.
Coupland, N. (2002) Language, situation, and the relational self: Theorising dialect-style in
sociolinguistics. In P. Eckert, and J. Rickford (eds.), Style and Sociolinguistic Variation. Cambridge:
Cambridge University Press, pp. 185-210. Style and social identities. Alternative approaches to linguistic
heterogeneity. Berlin: Mouton De Gruyter, pp. 213-46.
Coupland, N. (2007) Aneurin Bevan, class wars and the styling of political antagonism. In. P. Auer (ed.),
Style and social identities. Alternative approaches to linguistic heterogeneity. Berlin: Mouton De Gruyter,
pp. 213-46.
Deuchar, Margaret, Pieter C. Muysken, and Sung-Lan Wang (2007) Structured variation in
Codeswitching: Towards an empirically based typology of bilingual speech corpora. The International
Journal of Bilingual Education and Bilingualism 10.3: 298-340.
Dirven, R. (1992) Afrikaans: Feite en interpretasies. In V. Webb (ed.), Afrikaans ná Apartheid. Pretoria:
van Schaik, pp. 26-65.
Eckert, P., and J. Rickford (2002) Style and Sociolinguistic Variation. Cambridge: Cambridge University
Press.
Ervin-Tripp, Susan (2002) Variety, style-shifting and ideology. In P. Eckert, and J. Rickford (eds.), Style
and Sociolinguistic Variation. Cambridge: Cambridge University Press, pp. 44-56.
Esterhuyse, J. (1986) Taalapartheid en Skoolafrikaans. Emmarentia: Taurus.
Giliomee H. (2004) Die Afrikaners.’n Biografie. Capetown: Tafelberg.
Giliomee, Hermann (2003) The rise and possible demise of Afrikaans as a public language. PRAESA
Occasional Papers 14: 3-28.
Gumperz, J.J. (1982) Discourse strategies. Cambridge: Cambridge University Press.
446
Gerald Stell
Gumperz, J.J., and J. Cook-Gumperz (2007) A postscript: Style and identity in interactional
sociolinguistics. In P. Auer (ed.), Style and social identities. Alternative approaches to linguistic
heterogeneity. Berlin: Mouton De Gruyter, pp. 477-502.
Hopper, P.J., and E. Closs Traugott (2002) Grammaticalization. Cambridge: Cambridge University Press.
Labov, W. (1987) Are Black and White vernaculars diverging? American Speech 62.1: 3-66.
Lanham, L.W. (1978) An outline history of the languages of Southern Africa. In L.W. Lanham & K.P.
Prinsloo (eds.), Language and communication studies in South Africa. Cape Town: Oxford University
Press, pp. 13-28.
Malherbe, E. G. (1977) Education in South Africa. Vol. II: 1923-75. Cape Town: Juta.
McCormick, K. (2002) Language in Cape Town’s District Six. Oxford: Oxford University Press.
Meeuwis, M., and J. Blommaert (1998) A monolectal view of code-switching. Layered code-switching
among Zairians in Belgium. In P. Auer (ed.), Code-switching in conversation. Language, interaction and
identity. Cambridge: Cambridge University Press, pp. 76-98.
Muysken P. (1997) Code-switching processes: Alternation, insertion, congruent lexicalization. In M. Pütz
(eds.), Language choices. Amsterdam: John Benjamins Publishing Company, pp. 361-81.
Muysken, P.C. (2000) Bilingual speech: A typology of Code-Mixing. Cambridge: Cambridge University
press.
Orner J. (2008) Language policy and nation-building in Post-Apartheid South Africa. Amsterdam:
Springer.
Ponelis, F. (1993) The development of Afrikaans. Frankfurt am Main: Peter Lang.
Rickford, J. (2002) Style and stylising from the perspective of a non-autonomous sociolinguistics. In P.
Eckert, and J. Rickford (eds.), Style and Sociolinguistic Variation. Cambridge: Cambridge University
Press, pp. 220-234.
Roberge, P. (2004) Afrikaans. In A. Deumert & W. Vandenbussche (eds.), Germanic standardizations.
Amsterdam: John Benjamins Publishing Company, pp. 15-40.
Roefs, M. (2006) Identity and race relations. In U. Pillay, B. Roberts, & S. Rule (eds.), South African
social attitudes: Changing time, different voices. Cape Town: HSRC Press, pp. 77-100.
Scheffer, P. (1983) Afrikaans en Engels onder die Kleurlinge in die Kaapprovinsie en in besonder in die
Skiereiland [Afrikaans and English among the Coloureds in the Cape province and particularly in the
Peninsula]. Pretoria: HSRC.
Sebba, M., and T. Wootton (1998) We, they and identity. In P. Auer (ed.), Code-switching in
conversation. Language, interaction and identity. Cambridge: Cambridge University Press, pp. 262-286.
South African Statistics (2001) Census in brief. Pretoria.
Stell, G. (2009) Codeswitching and ethnicity: Grammatical types of codeswitching in the Afrikaans
speech community. International Journal of the Sociology of Language 199: 103-128.
Steyn, J.C. (1980) Tuiste in eie taal. Cape Town: Tafelberg.
Van der Ross, R.E. (1979) Myths and attitudes. An inside look at the Coloured people. Cape Town:
Tafelberg.
White and coloured identities in Afrikaans-English code-switching
447
Webb, V. (2002) Language in South Africa. The role of language in national transformation,
reconstruction and development (Impact. Studies in language and society). Amsterdam: John Benjamins
Publishing Company.
GERALD STELL is currently a Research Fellow for the F.W.O./V.U.B. He defended a Ph.D. thesis in
2008 on the topic of convergence and divergence between White and Coloured varieties of informal
spoken Afrikaans. His current research interests revolve around the role of ethnicity in grammatical
variation and code-switching mainly in the Southern African context.
Address: 23, rue Omer Lepreux, 1081 Koekelberg, Belgium. E-mail: [email protected]